Quantum cloud computing, delivered through the quantum-as-a-service (QaaS) model, gives users access to quantum processors without owning them. But pricing these services is tricky: a uniform time-based rate fails to account for the fact that quantum resources are fundamentally heterogeneous—different qubit architectures, error rates, and connectivity make some jobs more expensive to run than others.
The new paper, posted on arXiv, proposes using quantum reinforcement learning to orchestrate these resources. The goal is to find a scheduling policy that trades off cost and delay, rather than applying a one-size-fits-all pricing scheme. By framing the orchestration as a reinforcement learning problem, the approach can adapt to the varying characteristics of quantum hardware.
The work is early-stage, and the abstract does not provide experimental results. Still, it highlights a growing concern: as quantum clouds expand, pricing and scheduling will need to become more sophisticated than simple metered time.